Optimal Buffer Threshold Setting And Optimal Control Of Transmission Power Rate Based On VoD Video Streaming System

With the widespread popularity of smart mobile terminal equipment and services(such as Netflix,YouTube),the demand for video-on-demand(VoD)applications via the Internet is increas-ing.When users use the terminal to watch videos,they should need to send a request to the server,through the obtained video download address and initiate a resource request,the server sends the corresponding video streaming according to the requested information.However,as a large number of video streaming are transmitted through time-varying channel,and the time-varying fluctuation of available broadband,the video streaming flows into the playback buffer is extremely unstable,this phenomenon will affect the quality of service(QoS).In order to describe the impact of the video streaming playback process on the user,the International Telecommunication Union(ITU)has defined QoS of the system and quality of experience(QoE)for users.This work focuses on improving QoE for users.It discusses the optimal threshold setting of the video playback buffer in the wireless time-varying channel and the optimal control of the transmission power rate(TPR)of the video server.Due to the influence of factors,such as net-work congestion and natural environment,the channel gain of the wireless time-varying channel varies randomly.For the playback buffer,the video streaming's arrival rate is random,while the transcoding playback rate is fixed.This phenomenon will cause the video streaming system to hysteresis;But if a certain amount of video data can be buffered before the video is played,it will reduce the probability of hysteresis.Based on the stochastic fluid model(SFM)framework,this thesis proposes SFM model to investigate the wireless time-varying network dynamic video system.The relationship between the probability of hysteresis in the video streaming system and the initial buffering delay is studied,thereby the optimal buffer threshold setting is obtained.In addition,for the video streaming server,the changing channel gain and transmission power jointly determine the instantaneous throughput of wireless time-varying channels.Improper TPR settings will often cause the video stream to be unable to be transmitted within a given time or cause a lot of energy waste.By embedding the discrete decision time sequence,this thesis transforms the TPR control problem into a finite-stage discrete Markov Decision Prosses(MDP)with a mixed state and a finite set of actions,and gives the optimal TPR control strategy.In the setting of the optimal buffer threshold,this work firstly gives the three first passage times(FPTs)of the video streaming system,and derives the Laplace-Stieltjes transform(LST)matrix of FPTs.Based on this,the probability of the system hysteresis occurrence and the prob-ability of the initial buffer delay being higher than the given tolerance time are obtained.Then the algorithm for solving the optimal buffer threshold is given,and finally the theoretical results are verified by numerical analysis.In the TPR optimization control problem,This thesis estab-lishes a video streaming transmission model within a strict time limit,and proposes a problem of minimizing energy consumption based on this model,and transforms it into an MDP with a finite mixed state space(discrete state space and continuous state space).Further the continuous state space discretization,the transition probability is obtained through the Uniformization Algorithm,the above-mentioned model is constructed into an MDP sequence with only finite discrete states,and then the MDP sequence is solved to obtain the optimal strategy for TPR control.Finally,the numerical simulation verifies the validity and feasibility of related theories.